This initiative was carried out by a group of Mexican gastroenterologists interested in the subject. The coordinator of the group is an IBS expert (MS) and the participants were chosen based on their experience in gastroenterology and their training and participation in clinical and basic research related to the theme. An expert in the classification of levels of evidence and grades of recommendation (MP) with experience and training in gastroenterology, as well as clinical and statistical research, but who is not an IBS expert, was also included in the group. This was purposely done so that there would be both a different perspective and a more objective evidence evaluation. The 9 reviewers were divided into 6 reviewer groups (MB, RC-AH, ALC-JLT, YLV-MAV, MRT and MS), each receiving one of the 6 issues to be reviewed.

The project coordinator did a preliminary literature review in PubMed, using the MEDLINE database and including articles written up to July 2012. The following search terms were employed: «IBS» AND «SIBO», «abnormal breath test», «incidence of post infectious IBS», «prevalence of post infectious IBS», «microbiota», «Post infectious IBS» AND «risk factors», «epidemiology», «low grade inflammation», «Microbiota», «dysbacteriosis», «SIBO», «methane» AND intestinal function», «intestinal motility», «sensory function», «sensory abnormalities», «visceral hypersensitivity».

Once identified, the articles were distributed to those responsible for each theme to be reviewed. Systematic reviews with or without meta-analyses and original articles were selected. Narrative reviews were not included. In addition, the reviewers were authorized to include articles that were not selected in the initial review, but that were identified from other sources, such as from the references of an article originally chosen, or articles published after July 2012 and up to August 31, 2013, when the preparation of the manuscript concluded. All participants received a set of instructions by email with respect to the information to be obtained from the publications, as well as the methodology for classifying the levels of evidence and grades of recommendation.

The reviewers analyzed the evidence and elaborated statements based on the available information. The levels of evidence and grades of recommendation were evaluated and graded using the Oxford Centre for Evidence-Based Medicine (OCEBM) system.16 This system utilizes numbers and letters to evaluate the quality and the level of evidence of clinical studies. Quality and methodology are established with the numbers 1 through 5 and the lower case letter «a», «b», or «c». The numbers indicate the quality of the studies and the letters indicate the methodology employed. For example, a 1a study is usually a systematic review that only includes high quality, homogeneous, controlled clinical trials (the number 1 indicates that only high quality homogeneous controlled clinical trials were included and the letter «a» indicates that it is a systematic review); a 2 a study is a systematic review (letter «a») that includes cohort studies of different quality that are methodologically considered to be of lower quality and level (number 2) than the controlled clinical trials. A final example: 2 b is a single individual cohort or a single controlled clinical trial (letter «b») of low quality (number 2).

The grade of recommendation is given in the upper case letters A through D. The letter A is for statements, conclusions, or recommendations based on information obtained from high quality or level 1 evidence, whereas letter D is given to recommendations based on studies of low scientific quality or level 5 evidence.16

The first face-to-face meeting of the group was held in August 2012 and lasted 9hours. First the OCEBM system was discussed and then the reviewers presented a summary of each selected article in tables, including authors, journal, year, country, type of article (systematic review or original) and design, diagnostic criteria for IBS, and other selection criteria for the subjects, study methods and/or evaluated treatment, outcome variables, and results or conclusions. In addition, the reviewers responsible for each theme proposed a level of evidence for each of the studies and then presented the statements or declarations and their grades of recommendation. Each of the assigned levels of evidence was discussed and were modified and accepted by consensus; the same was done for the grades of recommendation. Finally, the coordinator presented a summary of each one and its pending work. In January 2013 the second meeting was held, lasting 8hours, and only the 6 updated reviews were presented. Then in March 2013 each reviewer sent his or her written participation to the coordinator who then sent each of the sections to be cross-reviewed. In other words, each reviewer or reviewer group went over another group's section. Once this cross-review was completed, the coordinator proceeded to edit the manuscript, after which it was reviewed again by all the participants.

• •

  Different studies have suggested that patients with IBS have a greater probability of having SIBO, determined through hydrogen breath tests (level 3 evidence, grade B recommendation).

• •

  The reported prevalence of SIBO in patients with IBS varies widely due to the different criteria for defining a positive breath test and the methodology employed (28 to 84% with the lactulose breath test [LBT], 2 to 31% with the glucose breath test [GBT], and 2 to 6% based on cultures) (level 3-4 evidence, grade C recommendation).

Twenty-four articles were identified that reported on the prevalence of SIBO in IBS; 3,11,17–38 23 articles during the initial search,3,11,17–35,37–38 and an additional article identified through manual search during the preparation of the document.36 Two systematic reviews with a meta-analysis that included more than 3,400 subjects and compared patients with IBS and healthy controls showed that the breath tests for SIBO were abnormal in the patients, with a 4-times higher probability than the controls.3,17 An extensive bibliographic search was carried out in both reviews, the studies were adequately selected, and the authors made a clear reference to the heterogeneity of the studies (Table 1).

On the other hand, a case series of patients with IBS that participated in an open study with rifaximin, showed a SIBO prevalence of 71% with LBT in IBS.37 In addition, 16 case-control studies11,18–30,36,38 provided information on SIBO prevalence in IBS, including the study by Pimentel et al.;18 the prevalence of this comparative controlled clinical trial with placebo was the result of a sub-analysis of the study population. A second study analyzed a series of consecutive patients with diverse digestive disorders that were referred for upper endoscopy; they had duodenal aspiration culture done to determine SIBO and IBS was considered a posteriori.36 In this same study SIBO was also compared in those patients with IBS-D and IBS-Non D, which made it incomparable with all the other studies.36 Of the 14 remaining studies, 5 demonstrated greater SIBO prevalence in IBS compared with the controls;3,17,19,24,38 7 showed equal prevalence,11,20–22,26–27,30 one showed lower prevalence in IBS,28 and one did not report the p value, although there appeared to have been a greater prevalence in IBS23 (Table 1). Among the analyses with greater prevalence in IBS, the study by Pimentel et al. stands out because it analyzed the prevalence of SIBO among patients with fibromyalgia, IBS, and healthy controls.19 The patients with fibromyalgia were selected regardless of their digestive symptoms and were the group with the highest SIBO prevalence, above that of the IBS patients and the healthy controls (100, 84, and 20%, respectively). It should be stressed that only one Latin American study was identified.21 In this study, Madrid et al., in Chile, found that the prevalence of SIBO was similar in patients with IBS, compared with that of other functional gastrointestinal disorders (IBS: 76%; functional constipation: 73%; functional diarrhea: 69%; and functional bloating: 68%).21 Another study analyzed patients treated with proton pump inhibitors (PPIs) vs those that were not, finding no apparent differences in SIBO frequency.25 And finally, two other studies compared IBS vs other functional gastrointestinal disorders (FGIDs).26,29 Six case series with a combined total of 478 patients were analyzed as well, and a prevalence of SIBO was reported in patients with IBS that varied from 36 to 74%, depending on the methods employed.31–35,37

The use of breath tests for diagnosing SIBO has been characterized by the lack of a standardized methodology and validated criteria for defining an abnormal test. In the majority of the studies lactulose was the substrate, but with a wide variety of doses, protocols for carrying out the tests, and criteria for determining that a test is abnormal. Walters et al. used the LBT and applied 2 different criteria in their interpretation.20 They included 39 patients with IBS and 20 healthy controls, finding a radically different SIBO prevalence in patients with IBS, even though there was no difference in the comparison with the healthy controls, regardless of the criterion used: 28% of the patients with IBS vs 30% of the control subjects when the criterion of more than 20ppm of H2 in the first 90min of the test was used, compared with 69% of the patients with IBS vs 75% of the controls when the criterion of more than 20ppm of H2 at any time during the first 180min of the test was employed.20 Furthermore, the accuracy of the LBT has been questioned because the decomposition of this substrate by the bacteria of the cecum usually produces a second spike in hydrogen detection that reduces its specificity. In contrast, the GBT, in which the substrate is completely absorbed in the proximal small bowel (duodenum), has shown a greater sensitivity and specificity for the detection of SIBO than the LBT.11 Rana et al. found a similar SIBO prevalence in IBS patients and healthy subjects utilizing the LBT (34 vs 30%, p = NS), but a higher prevalence of SIBO in IBS utilizing the GBT (6.2 vs 0.66%, p<0.01).30 Despite its apparently being a more precise test, it has been employed in fewer studies.24,27,32,36 Sucrose, another substrate that is totally absorbed in the small bowel and therefore theoretically more accurate, was used in only one of the selected studies.23

On the other hand, the presence of SIBO has also been defined based on the detection of an elevated bacterial count from small bowel fluid culture. The quantitative culture should then be regarded as the benchmark test for SIBO, but it has been utilized in very few studies. However, it would depend on the culture site and many bacteria are unculturable. Only 2 of the selected studies employed culture for detecting SIBO,22,28 and as mentioned before, one of them reported a lower SIBO frequency in IBS vs control subjects with different pathologies that underwent endoscopy.28

We can therefore conclude that there is evidence suggesting a greater probability of SIBO in IBS according to breath test data, but there is not enough evidence for recommending the routine use of these tests for diagnosing SIBO in IBS.

• •

  The composition of the microbiota in patients with IBS is different from that of normal subjects (level 3 b evidence, grade B recommendation).

• •

  Alterations in the composition of the microbiota -dysbiosis- occur in both adult and pediatric patients with IBS (level 3 b evidence, grade B recommendation).

• •

  Due to the heterogeneity of IBS and the use of different methods for studying the gut microbiota, it is not possible to establish a microbial composition characteristic of IBS (level 3 b evidence, grade B recommendation).

Twenty-six published articles were identified that studied the composition of the microbiota in patients with IBS; 24 in the initial search41–64 and 2 from other sources.39,40 All of them are case-control studies conducted in Europe, Asia, and the United States; none from Latin America or Africa. Twenty-five were carried out on adult population39–55,57–64 and only one on children.56 In 11 of the studies, the cases were classified according to the IBS subtype.41,43,46–48,52,55–56,60,62,64 The microbiota was analyzed with molecular methods in the majority of the studies, whereas just fecal culture was used in 2,39,40 and both methodologies were used in 4 studies.42,48,51,58 Even though the majority of the studies analyzed the composition of the microbiota in samples of fecal matter, the microbial composition was also examined in biopsies of the colonic mucosa in 4 of the studies. The predominant results of each of these studies and the summary of the microbial ecology of the gut microbiota in IBS are shown in Table 2.

The investigations that used fecal cultures for studying the gut microbiota have shown that IBS patients, unlike healthy subjects, have a diminished population of bifidobacteria and lactobacilli and an increased population of streptococci, coliforms, and Clostridium species.39,40,42,51 Moreover, the majority of the studies used molecular methods independent of the culture, such as tests based on DNA extraction and amplification of the 16S genes of ribosomal RNA, quantitative PCR, the products of PCR through denaturing gradient gel electrophoresis, and probe-specific fluorescence in situ hybridization. The many different molecular strategies employed in these studies (Table 3) is the reason for the inconsistent and even contradictory results in relation to the composition and diversity of the microbiota in patients with IBS, as well as a single determination of the microbiota in the variable of time and the limited knowledge of new bacterial species that are still waiting to be described. Thus, even though it seems that the gut microbiota of patients with IBS is different from that of the controls, it is not yet possible to establish an intestinal microbial composition characteristic of IBS.

• •

  The average incidence of PI-IBS has been reported as 9 to 10% with a 4 to 36% interval (level 1 a evidence, grade A recommendation).

• •

  The prevalence of PI-IBS varies from 3 to 17% and decreases over time after gastrointestinal infection (level 3 b evidence, grade B recommendation).

• •

  The most studied etiology in relation to PI-IBS is that of bacterial origin, and even though the viral and parasitic causes have scarcely been studied, they also appear to be risk factors for PI-IBS (level 2 b evidence, grade B recommendation).

Twenty-three studies on PI-IBS were reviewed, 19 of which were identified in the initial search10,65–82 and 4 afterwards from other sources.83–86 Twelve studies reported the incidence of PI-IBS (new onset IBS),65–67,71–72,75,77,80,83,85 8 reported the prevalence,65,69,73,74,78,79,82,84 and 8 analyzed the risk factors related to the development of PI-IBS.10,66,70,72,82,84,86 All the studies were conducted on adult population, with the exception of one on pediatric population71 (Table 4).

The incidence of clinical symptoms of IBS after a gastrointestinal infection has been reported at an average of 9-10% based on 2 systematic reviews, but varies depending on the case from 4 to 36%.65,67 There are no differences if IBS develops after an acute gastroenteritis episode during epidemics, due to isolated infections, or after traveler's diarrhea.65 Likewise, the probability of developing IBS is 6 times higher in subjects that have been exposed to gastrointestinal infections than in those that have not.66

The prevalence of PI-IBS has been reported in 7 to 33% of patients, but there are wide variations depending on the reported series and particularly on the time of observation.65 Prevalence also varies depending on the geographic region and Mexico appears to have one of the lowest prevalence rates in the world, at only 5.0%.79 Prevalence is also higher if it is evaluated sooner rather than later after an infectious outbreak. For example, 2 years after an outbreak of bacterial gastroenteritis in Walkerton (Canada), PI-IBS prevalence was reported in 30.4% of the subjects exposed to acute gastroenteritis.87 Contrastingly, in the following years the prevalence had decreased and at 8 years it was 15.4%.70,88 Similarly, in Sweden the initial PI-IBS prevalence of 12% was reduced to 9%, 5 years later.84 In large reviews it has been reported that the probability (odds ratio: OR) of developing IBS 3 months after an episode of infectious diarrhea was 7.58-8.47 times higher than in the control population, but at 24 to 36 months the OR had descended to 3.85-4.05.10,66

Regarding the causal agent of PI-IBS, the studies on incidence and prevalence generally refer to clinical presentations of IBS after bacterial infections or the cause is not specified. The most frequently identified bacteria have been E. coli, Campylobacter, Shigella, and Salmonella.10,67,68,77,80E. coli was the cause in the majority of the patients presenting with PI-IBS after an episode of traveler's diarrhea acquired in Mexico.80 In a group of patients in Houston (Texas) 16.1% of the patients with PI-IBS had previously travelled abroad, whereas only 7.5% of the patients with non PI-IBS had done so.69 A study conducted on children reported PI-IBS in 10.5% after Campylobacter infection, compared with IBS in 2.5% of the children that were not exposed.71 On the other hand, bacterial gastroenteritis due to Campylobacter is followed by IBS more frequently than by infectious diseases that do not affect the digestive tract, such as infectious mononucleosis, for example.81 With respect to gastroenteritis of viral etiology, Norovirus has been described as causing PI-IBS; the results of the 2 published studies on this75,85 coincide, reporting that 21.5 and 23.6% of the patients had PI-IBS, whereas only 1.5 and 4.4% of the controls had IBS. In relation to the role of intestinal parasites, the results are less conclusive. A Central American study found no differences in IBS prevalence according to the Rome II criteria in individuals with a history of parasitosis vs subjects with no such history (16.6% vs 15.4%).73 On the other hand, after a giardiasis outbreak that infected a large number of Norwegians, the prevalence of IBS according to the Rome III criteria was noticeably higher than in the control population (46 vs 14%).74 Likewise, in an outbreak of Trichinella britovi in Turkey that resulted in 72 cases of infection, 10 developed IBS (13.9%).83

In reference to the risk factors for developing PI-IBS, the female sex, the severity of gastroenteritis, and the presence of anxiety and depression have been described.10 Villani et al. analyzed the subjects that developed PI-IBS 2 to 3 years after the Walkerton epidemic, and found that genetic variations associated with the expression of the Toll-like receptor (TLR)-9 related to innate immunity, interleukin (IL)-6 associated with immune activation, and cadherin-1 (CDH1) involved in tight epithelial junctions, were independent risk factors for PI-IBS. 76

The above allows us to conclude that the incidence and prevalence of PI-IBS are variable and even though the bacterial etiology has been studied the most, it appears that viruses such as the Norovirus and parasites such as Giardia may also be related to PI-IBS. In addition, risk factors such as the female sex, severity of gastroenteritis, and previous anxiety and depression, as well as genetic factors associated with immunity, have been determined.

• •

  There is evidence that suggests the presence of low-grade intestinal inflammation in a subgroup of IBS patients, which involves an increase in intraepithelial T lymphocytes (IEL), mast cells and enterochromaffin cells (level 3 a evidence, grade B recommendation).

• •

  The increase in IEL and mast cells appears to be more commonly observed in patients with IBS-D, compared with IBS-C and IBS-M; however, whether there are differences between PI-IBS and non PI-IBS cannot be concluded (level 3 a-b evidence, grade B recommendation).

• •

  There is insufficient evidence to determine whether there are differences in the enterochromaffin cells between PI-IBS and non PI-IBS (level 5 evidence, grade D recommendation).

A total of 29 articles were identified; 2 were systematic reviews 89,90 and the rest were original ones.91–117 Twenty-seven studies were identified in the initial search90–115,117 and 2 89,116 were later selected from other sources. All the studies were conducted on adult population, with the exception of one on pediatric population. Twenty-four studies analyzed the presence of chronic inflammatory cells (T lymphocytes, mast cells, and enterochromaffin cells) in the mucosa of the colon and rectum in IBS patients and controls89,91,94–104,107–108,110–118 (Table 5).

For several years there have been reports on the increase in the number of enterochromaffin cells in rectal biopsies of PI-IBS patients.99–100,106 Spiller et al. reported an up to 5 times higher increase in the number of enterochromaffin cells positive for synaptophysin in patients with C. jejuni infection.95 A gradual decrease in the number of enterochromaffin cells was observed in these patients in biopsies taken 6 and 12 weeks after infection; however, one year after the acute infection in the subgroup of patients that remained symptomatic, that is, those with PI-IBS, the number of enterochromaffin cells remained elevated, in a range similar to that observed 2 weeks after the C. jejuni infection. The higher number of enterochromaffin cells may have pathophysiologic importance because these cells are the main source of serotonin (5-HT) storage in the organism and there is evidence of an increase in 5-HT release in IBS patients.119–120 The prokinetic and secretory effect of 5-HT may be related to the diarrhea or liquid stools that accompany IBS-D. A recent systematic review89 concluded that despite the fact that some researchers have observed an increase in the number of enterochromaffin cells and in the production of serotonin in the mucosa of the colon and rectum in IBS patients, compared with healthy controls, many others have not confirmed such findings. The results show that these changes are not consistent.

In addition, some studies have demonstrated a rise in the number of IEL in both IBS-D and PI-IBS, mainly after acute gastroenteritis due to C. jejuni or Shigella.95,111,117 Nevertheless, it is not completely known if there is also an increase in T lymphocytes in non PI-IBS. In fact, only 7 studies compare PI-IBS and non PI-IBS with respect to the inflammatory changes encountered through histology.92,94,99–100,108,110–111 Dunlop et al. found a higher number of enterochromaffin cells and IEL in PI-IBS than in non PI-IBS and the controls in 2 studies, and therefore suggest that they could be markers for PI-IBS.99–100 Likewise, Lee et al. observed a greater number of enterochromaffin cells, IEL, and mast cells in rectal biopsies in PI-IBS patients, compared with non PI-IBS patients and healthy controls.108 An increase in the number of mast cells in non PI-IBS was observed only in those patients with IBS-D, not in patients with IBS-C or IBS-M. The rise in the number of mast cells had been previously described by Weston et al. in biopsies of the terminal ileum in patients with IBS, compared with the control group, but no differentiation was made between PI-IBS and non PI-IBS.93 Other researchers later confirmed the increase in the number of these cells in IBS,96,98,101,106 mainly in the IBS-D subgroup, in the patients with PI-IBS, as well as in those with non PI-IBS. Furthermore, the mast cells95–96,98,101,106 appear to be near the sensory neurons, and there is a positive correlation with the severity and frequency of pain and/or abdominal discomfort when they are in closer proximity. 101

In contrast, Braak et al. reported a decrease in the IEL, macrophage, and mast cell count in the colonic mucosa in 66 patients with IBS, compared with 20 healthy controls.115 In that study, the difference between PI-IBS and non PI-IBS was not specifically analyzed, but rather the patients with acute and gradual IBS onset were compared, and a lower number of macrophages was observed in those with gradual IBS onset.115 It is likely that the sudden onset group corresponds to PI-IBS, but we cannot conclude that. Previously, another study by the same group in Holland not only found a lower number of mast cells in biopsies of the rectum and descending colon in IBS, but also reduced tryptase release, compared with the controls.91 Finally, Chang et al. found no differences in the number of immune cells in the colonic mucosa between patients with non PI-IBS and the controls.116

The above suggests that there is an increase in IEL, mast cells, and enterochromaffin cells in the intestinal mucosa in a group of patients with IBS that appears to be more frequent in those with IBS-D. However, it cannot be determined whether this low-grade inflammation is characteristic of PI-IBS or non PI-IBS.

• •

  The evidence suggests that the differences in the composition of the microbiota in subjects with IBS are related to alterations in the visceral sensitivity and motility function of the gastrointestinal tract (level 1 b evidence, grade A recommendation).

• •

  The presence of methanogenic microbiota is significantly associated with constipation predominant IBS (IBS-C) (level 3a evidence, grade B recommendation).

Eight articles related to bowel function22–23,35,60,94,121–123 were identified in the initial search and 3 additional articles104,124–125 were identified from other sources (Table 6). The evidence suggests that the changes in the microbiota of the patients with IBS have an influence on visceral sensitivity and gastrointestinal motility, especially at the antroduodenal and colorectal level.22–23,35,60,94,121–126 Regarding the sensory distubances, the studies have shown that some patients with IBS and dysbiosis (PI-IBS and IBS with SIBO) develop rectal hypersensitivity, one of the most characteristic pathophysiologic findings in IBS.94

Various studies have also described that patients with PI-IBS have faster colonic transit time. For example, Gwee et al. demonstrated that subjects with a history of gastroenteritis and IBS had faster colonic transit than a control group of healthy subjects (median colonic transit time of 34.4 vs 55.2min, p=0.01).94 In contrast, Yu et al. found that the orocecal transit time correlated with the IBS subtype.35 Thus, orocecal transit is longer in patients with IBS-C than in those with IBS-D (p=0.0023).

Moreover, in patients with IBS and evidence of SIBO alterations in the number and frequency of phase III activities of the migrating motor complex (MMC) have been described.22–23,121 For example, in a group of patients with IBS according to the Rome I criteria and SIBO based on a positive LBT, Pimentel et al. demonstrated that the number of events of phase III of the MMC was lower in patients with IBS and SIBO than in the healthy controls (0.7 vs 2.2, p<0.001); the same was true for the phase III duration (305 vs 428 s, p<0.001).121

Numerous studies have suggested that patients with IBS have qualitative changes in the colonic flora. For example, there are descriptions of patients that can develop a proliferation of bacterial species that produce more gas, specifically methane. The presence of methanogenic flora in patients with IBS has been associated with a slower colonic transit time, rectal hyposensitivity, and altered intestinal motility.122–125 In a recent analysis of the microbiota in patients with IBS through the technique of pyrosequencing, Jeffery et al. demonstrated that 17 taxas are associated with a slow colonic transit time (including those of the following phylotypes: Euryarchaeota, those of the class: Methanobacteria, and those of the families: Methanobacteriaceae and Desulfohalobiaceae).60 Likewise, the presence of Proteobacteria is described as being associated with an increase in the pain threshold during rectal distension, evaluated using a barostat. In addition, the first evidence in IBS of the detection of different levels of acetic acid, propionic acid, and total fatty acids has been reported. The highest levels are associated with poor outcome IBS.

In summary, the evidence suggests that changes in the composition of the microbiota or its instability (dysbiosis) have an influence on the gastrointestinal physiology, producing abnormalities in visceral sensitivity and gastrointestinal motility. However, further studies are required in order to determine the effect of the microbiota on those sensory and motor alterations. On the other hand, it is not known whether these disturbances contribute to symptom generation or whether they are the consequence of primary motility disorders. Finally, it should be stressed that there are other factors that have an influence on the microbiota of patients with IBS, such as the type of diet (i.e. FODMAPs: fermentable oligosaccharides, disaccharides, monosaccharides and polyols) and the use of antibiotics.

• •

  In patients with non-constipation IBS, rifaximin at doses of 400mg TID for 10 days or 550mg TID for 14 days, is superior to placebo in the adequate response of global IBS symptoms and in abdominal bloating. It also improves pain and abdominal discomfort, as well as the consistency of loose/liquid stools during treatment and for up to 10 weeks post-treatment (level 1 b evidence, grade A recommendation).

• •

  Rifaximin at a dose of 400mg TID for 7 days may neutralize the LBT in approximately half the patients with IBS, which is associated with reduced IBS symptom severity (level 4 evidence, grade C recommendation).

• •

  The frequency of adverse events is similar between rifaximin and placebo, and the most frequent are: headache, upper respiratory tract infections, nausea, nasopharyngitis, diarrhea, and abdominal pain (level 1 b evidence, grade A recommendation).

• •

  In the patients that require retreatment with rifaximin, effectiveness appears to be similar to that of the first treatment; however, further studies are required in order to determine the effectiveness of retreatment and the appropriate interval for carrying it out (level 4 evidence, grade C recommendation).

• •

  Studies are required for evaluating the long-term safety and effectiveness of rifaximin in IBS (level 5 evidence, grade D recommendation).

Twenty articles that analyzed antibiotic therapy in IBS were identified, 5 reviews 15,127–130 and 15 original articles,18,33,131–143 all studies on adults, except one conducted on the pediatric population.135 An additional original article that had been published in December 2013 was added 37 (Table 7).

Basically 2 antibiotics have been studied in IBS: rifaximin and neomycin. Rifaximin is a semi-synthetic antibiotic, an analog of rifamycin, designed to have little gastrointestinal absorption. It inhibits the bacterial synthesis of RNA by binding to the RNA polymerase β subunit that is dependent on bacterial DNA.144 Its absorption is less than 0.4%, making it an almost completely luminal-acting antibiotic, and most of it is excreted in the fecal matter, unchanged. It has a broad spectrum of activity against Gram-positive, Gram-negative, aerobic, and anaerobic enteropathogens145–146 with a low probability of bacterial resistance.147–148 In vitro it induces CYP3A4, but its interaction with other medications is almost null.129

In the selected reviews,15,127–131 a short cycle of non-absorbable antibiotics, like rifaximin, has been recommended for improving global IBS symptoms.131 The authors of a systematic review with only 3 original articles on rifaximin treatment of IBS with or without SIBO concluded that rifaximin at a dose of 400mg BID for 7 to 10 days, improves IBS symptoms, regardless of whether SIBO is present or not.129 In 2 later systematic reviews with meta-analyses published 2 years apart, the second one with 10 times more patients treated with rifaximin vs placebo, it was concluded, and with a good level of evidence, that the antibiotic at doses of 400 to 550mg, 2 to 3 times/day, was twice as effective as the placebo at improving IBS symptoms.15,127 Additionally, rifaximin showed a therapeutic gain of 9.8 over the placebo and a number needed to treat (NNT) of 10.2 for global IBS improvement and very similar values for abdominal bloating improvement.15,127 In another systematic review on the treatment of abdominal bloating, it was concluded that rifaximin was superior to placebo in the proportion of patients with IBS-Non C that reported subjective improvement in abdominal bloating.129

The original articles analyzed rifaximin and neomycin, as well as a few other antibiotics. However, the studies have different designs and include retrospective studies, case series, and randomized controlled trials with placebo or other antibiotics and different doses. For example, rifaximin has been studied from doses of 200mg QID,137 400mg BID or TID for 7 to 14 days, to 550mg TID for 14 days.15 Likewise, the outcome variables analyzed were very different in each of the studies, from global IBS improvement to improvement of secondary symptoms such as pain and/or abdominal bloating, to the frequency and consistency of bowel movements.131,132,134 The effect of rifaximin on GBT136,137 or LBT33 has been evaluated as well. These differences in the studies make it difficult to arrive at a conclusion. Nevertheless, the best evidence comes from the recent Target 1 and Target 2 studies by Pimentel et al., with more than 1,200 patients between the 2 investigations.134 They show that rifaximin at a dose of 550mg TID for 14 days in patients with IBS-Non C by Rome II criteria resulted in a greater proportion of patients with adequate relief from IBS symptoms, abdominal bloating, daily symptom intensity, abdominal pain, and stool consistency.134 In the follow-up at 10 weeks post-treatment, adequate relief defined as the subjective report of symptom improvement in at least one of every 2 weeks, rifaximin remained significantly superior to placebo. It is worth noting that it had previously been reported that at lower doses, such as 400mg BID or TID for 10 days, rifaximin was also superior to placebo in the percentage of patients that reported overall improvement of IBS.131–132 Regarding secondary effects, in the Target 1 and 2 studies, which are the largest, frequency was similar to the placebo (1.6% with rifaximin and 2.4% with placebo) and the main effects in order of frequency were headache, followed by respiratory tract infections, abdominal pain, nausea, and diarrhea.134

Furthermore, rifaximin can neutralize LBT in 52% of the patients with Rome II-IBS and in this subgroup symptom severity improved, but the corresponding study was not controlled,33 and neither were those that evaluated the effect on GBT.136,137 Regarding the effect on children, a dose of 550mg TID for 10 days showed no differences from the placebo or in LBT normalization, which was achieved in only 20%.135 This suggests that children most likely require higher doses or a different type of antibiotic, probably due to a more resistant microbiota. In reference to rifaximin retreatment, only one retrospective study analyzed this modality; patients treated up to 5 times had a mean interval of symptom recurrence of 4 months.143 Effectiveness was 75%, similar to that of the first treatment. However, well- designed studies are required in order to determine how often rifaximin can be repeated.143 Finally, no studies have evaluated the long-term effects of rifaximin in IBS.

Regarding neomycin, it is a systemic aminoglycoside that has been evaluated in 2 original studies on IBS,18,138 both of which are randomized and controlled with placebo. At a dose of 500mg BID for 7 to 10 days in patients with IBS in general in one study, and IBS-C by Rome I criteria in the other, neomycin was superior to the placebo in the percentage of patients, mainly those with a positive LBT, that reported improvement in a composite score of symptoms including abdominal pain, diarrhea, bloating, and bowel habit.18 Neomycin normalized LBT in only 20% of the patients, but in the methane-positive patients, the percentage that reported improvement in regard to constipation was 9 times higher with neomycin than with placebo.138 Despite this, neomycin is not an ideal drug for IBS because of its characteristics in relation to systemic absorption and its safety spectrum.

With regard to other antibiotics, a retrospective study was conducted that compared rifaximin with neomycin, doxycycline, and amoxicillin-clavulanic acid that were used in the management of the treatment and retreatment of SIBO in patients with IBS;141 however, these agents were not as effective as rifaximin. Erythromycin has also been studied, evaluating the number of days until IBS symptom recurrence after LBT neutralization, but it was much less effective than tegaserod.142 And finally, rifaximin together with metronidazole has been evaluated in post-giardiasis-IBS, but no conclusion can be reached in relation to this study.139